a product life cycle ontology for additive manufacturing · the supplier, shop, machine, device,...
TRANSCRIPT
A Product Life Cycle Ontology for AdditiveManufacturing
Munira Mohd Alia,∗, Rahul Raia, J. Neil Otteb, Barry Smithb
aDepartment of Mechanical and Aerospace Engineering, State University of New York atBuffalo, Buffalo, NY, USA
bNational Center for Ontological Research, State University of New York at Buffalo, Buffalo,NY, USA
Abstract
The manufacturing industry is evolving rapidly, becoming more complex, more
interconnected, and more geographically distributed. Competitive pressure and
diversity of consumer demand are driving manufacturing companies to rely more
and more on improved knowledge management practices. As a result, multiple
software systems are being created to support the integration of data across
the product life cycle. Unfortunately, these systems manifest a low degree of
interoperability, and this creates problems, for instance when different enterprises
or different branches of an enterprise interact. Common ontologies (consensus-
based controlled vocabularies) have proved themselves in various domains as a
valuable tool for solving such problems. In this paper, we present a consensus-
based Additive Manufacturing Ontology (AMO) and illustrate its application in
promoting re-usability in the field of dentistry product manufacturing.
Keywords: additive manufacturing ontology, manufacturing process
ontology, ontology engineering, product life cycle, dentistry product
manufacturing
IPreprint version of article published in Journal of Computers in Industry, Volume 105,February 2019, Pages 191-203. (https://doi.org/10.1016/j.compind.2018.12.007) .
IIOnly to be used for personal use, copyright belongs to the authors and Elsevier Science.∗Corresponding authorEmail addresses: [email protected] (Munira Mohd Ali), [email protected]
(Rahul Rai), [email protected] (J. Neil Otte), [email protected] (Barry Smith)
Preprint version of paper published in Computers in Industry, 105 (2019), 191-203https://doi.org/10.1016/j.compind.2018.12.007
1. Introduction
The economic model of the manufacturing industry is increasingly based on
the modularization of industrial processes and digitally mediated collaboration
between modules both internally, within the enterprise, and externally, through
subcontracting and off-shoring. The sharing of information is crucial for facilitat-5
ing such collaboration across all phases in the life of a product from development
and design, through production and sale, to use and disposal [1, 2, 3]. As
industrial processes come to be further modularized and distributed throughout
the entirety of the manufacturing pipeline, more powerful and more intelligent
software solutions are required to support the different components and phases10
of the product life cycle (PLC).
The Economist Intelligence Unit [4] reports that the need for knowledge
representation of manufacturing processes is increasing exponentially as tech-
nology expedites the rapid exchange of information. In information science, an
ontology is a controlled vocabulary implemented in a semantic or knowledge15
representation language such as the Web Ontology Language (OWL). Ontologies
have been successfully used, for example, in military domains and biomedicine.
In the design and manufacturing domains, in contrast, the use of ontologies
has not lived up to initial expectations, due not least to a lack of coordination
among industrial enterprises. The focus of this paper is to present the Additive20
Manufacturing Ontology (AMO), an ontology designed to represent the Additive
Manufacturing (AM) Product Life Cycle.
AMO is a modular ontology that employs Basic Formal Ontology as its
top level, while also drawing from the Common Core Ontologies and three
other ontologies from the Coordinated Holistic Alignment of Manufacturing25
Process Ontologies that represent manufacturing processes: the Manufacturing
2
Process Ontology, the Design Ontology, and the Testing Process Ontology. As an
illustrative example, the use of AMO is illustrated in the dentistry manufacturing
domain.
2. Related Work30
2.1. Ontology Development in the Manufacturing Domain
Many manufacturing ontologies have been developed in recent years. For
example, Lemaignan et al. [5] developed the Manufacturings Semantics Ontology
(MASON), which employs three top-level classes of entities, operations, and
resources. Entities in MASON comprise a broad class including geometric entities35
(for example, shape), raw materials, and costs. Operations class attempts to
cover all processes involved in manufacturing, and resources attempt to represent
tools, human resources, and geographic resources. MASON was developed as an
upper-level ontology to accomplish two goals:
1. Developing an architecture and tools for automatic cost estimation, and40
2. Linking a high-level ontology written in OWL with a multi-agent framework
for manufacturing simulation.
Unfortunately, MASON’s tripartite division of classes into entities, operations,
and resources are lack of classificatory coherence. Entity in MASON, for instance,
is introduced as comprising the common helper concepts used to specify a product.45
However, entity as defined in the OWL Web Language Guide [6] and OWL2
Web Ontology Language Structural Specification [7] does not limit the definition
in specific concept of specifying product only but also include classes, datatypes,
object properties, data properties, annotation properties, and named individuals
are entities.50
Kjellberg et al. [8] introduces the Machine-Tool Model (MTM) as an ontology
focusing on the machine tool as a central part of a manufacturing system as
3
well as on the way machine tool information is used throughout the design and
operation of such systems. Process planning, for instance, requires information
on the functional properties of machine tools, such as the ability to perform55
different types of machining operations.
Both MASON and the MTM were developed from scratch, each in its ad hoc
way, and they do not use a common upper-level ontology nor do they reuse the
content of other domain ontologies. In this way, they re-create the very lack of
interoperability that they were designed to address, but now this lack occurs60
between ontologies rather than between data systems. Even though MASON was
intentionally developed as an upper-level ontology to represent manufacturing
information, the entities in MASON are identifiable and concrete to represent
manufacturing as the specialized domain of interest instead of being an ontology
that is domain neutral. According to Musen [9], upper-level ontology is defined65
as an ontology at a sufficiently high level of abstraction such that it does not
refer to identifiable, concrete entities in the domain of interest.
MASON’s contribution as an upper-level ontology is undeniably has con-
tributed to the development of other ontologies in manufacturing domain. For
instance, Ramos [10] introduces the Machine Ontology (MO), which elaborates70
the representation of machines in terms of the market, material, and operation
features. The resultant redundancy between MASON, MTM, and MO led Ramos
et al. [11] to present a method for integrating ontology reuse with ontology vali-
dation, and they applied this method to the three ontologies in question, using
Protege-Prompt to find common content and overlapping terms between them.75
The Machine of a Process Ontology (MOP) was developed as a result of this work
with the goal of facilitating the buying and selling of industrial machinery; it
employs MASON as its reference ontology while drawing in relevant classes and
relations from MTM and MO. The CDM-Core Ontology, presented by Mazzola
4
et al. [12] is another ontology that was developed by reusing MASON as one80
of the upper-level ontology. CDM-Core Ontology includes both the general
manufacturing domain applicability and the specific project use cases that can
be a guidance for developing other specific applications in manufacturing domain.
Another ontology, the Manufacturing Service Description Language (MSDL),
was introduced by Ameri et al. [13], who employ a methodology relying on85
the incremental enhancement of an initial set of definitions constructed on the
basis of a formal ontology. MSDL is an upper-level ontology that supports
the semantic framework for representing conventional manufacturing processes
outlined in Kjellberg et al. [8]. The original purpose of MSDL was to serve as
the ontology in an agent-based framework for supply chain deployment; for this90
reason, it employs an analysis of manufacturing capabilities across several levels:
the supplier, shop, machine, device, and process.
Mesmer and Olewnik [14] proposes a Part-Focused Manufacturing Process
Ontology (PMPO) designed around the idea that a classification of manufacturing
processes can be developed on the basis of an account of the desired features and95
attributes of the products they will be used to manufacture. The ontology thus
develops a representation of the qualities used in specifying product requirements,
including material composition, cost, shape, size, the surface finish of the product,
thickness, and so forth. Users can describe the features and attributes based on
the qualities defined in PMPO, and select appropriate manufacturing processes100
according to the information provided.
Most ontologies designed for the manufacturing domain thus far have been
put together with a focus narrowly directed to some specific sub-domain of
manufacturing engineering and with little attention to interoperability with
other ontologies in related domains. Among the ontologies discussed, MOP105
and PMPO stand out because they build on prior work. PMPO is especially
5
interesting in that it utilizes not only Basic Formal Ontology but also MSDL,
the Ontology for Biomedical Investigations (OBI), and the Common Semantic
Model Ontology (COSMO). It is thus, at least to some degree, able to achieve
interoperability among data systems deriving from external sources. On the110
other hand, PMPO is small in scale and has been designed only for traditional
machining and molding processes thus it cannot be applied to more modern
manufacturing processes such as additive manufacturing.
Ideally, a representation of the manufacturing domain should deal with
commonly collected product-related information. Moving forward, we hold115
that an ontological representation of products and the PLC is a prerequisite
for integrating data across systems in the manufacturing domain. Therefore,
developing an ontology with a focus on AM products - their qualities, functions,
the production, use, and end-of-life is the main objective of this paper. However,
our ontology is intended to form part of a larger suite of modular ontologies120
within the framework of the Industrial Ontologies Foundry (IOF) initiative, and
it will accordingly be modified in tandem with IOF development. IOF is an
initiative that was proposed to promote interoperability of high quality and
non-redundant ontologies in industrial domains or manufacturing specializations
[15].125
2.2. Manufacturing Processes and the Product Life Cycle (PLC)
As customer demands diversify, the complexity of products and product
repertoires increases, and this gives rise to demand for increasingly innovative
manufacturing processes [16]. Understanding the nature of such processes and
creating computational systems that can understand and reason about them is130
crucial, and this means understanding and reasoning across the entire product
life cycle (PLC) in the Product Life Cycle Management (PLM). Moreover,
Product, Process and Resource (PPR) are the key elements of engineering
6
domain in any manufacturing industry [17]. The information about PPR that
is structured in PLM systems requires explicit mapping among the PPR for135
a complex decision purpose. Therefore, an ontology that provides common
vocabularies in representing knowledge could facilitate the full potential of the
PLM by supporting information exchange between the PPR in different phases
in PLC[17, 18].
Cao and Folan [19] divide PLC models into two groups:140
1. Marketing Product Life Cycle (M-PLC) models that focus primarily upon
marketing needs and conceptions;
2. Engineering Product Life Cycle (E-PLC) models that integrate design and
manufacturing with marketing needs and conceptions.
Figure 1 represents the successive phases of the PLC taken as the basis of145
ontology development in many recent works, including Young et al. [3], Chen
et al. [16], Borsato [18], Matsokis and Kiritsis [20], Chungoora et al. [21], Usman
et al. [22], Urwin [23], Urwin et al. [24], and others.
Figure 1: Phases in the PLC.
Table 1, from Chen et al. [16], extends this representation to create a more
granular perspective. Here, the PLC is depicted as consisting of seven stages150
each with a number of sub-stages.
Chen et al. [16], Borsato [18], Matsokis and Kiritsis [20], Chungoora et al.
[21], Usman et al. [22], Urwin [23] and Urwin et al. [24] have demonstrated
good concept of information integration and sharing between the design and
manufacturing phases in PLC through ontology. Borsato [18] for instance,155
7
Table 1: Seven stages of the PLC presented by Chen et al. [16]
Stages Sub-stagesProduct design Requirement analysis, Conceptual design, Prelimi-
nary design, Detail designProcess development Part description, Generative process planning, Vari-
ant process planningProduct manufacturing Equipment layout, Production management, Qual-
ity controlSales Chance analysis, Target market choice, Sell combi-
nation developmentProduct in use Operation instructions establishment, Product in-
stallation and executionPost sell service User problem reaction, Problem identification, Ser-
vice supportProduct retirement Decomposition, Recycling
believes that ontology could bridge the gap between manufacturing and PLC.
Chungoora et al. [21] presents the Interoperable Manufacturing Knowledge
System (IMKS) model-driven concept that was built on the ideas of extensible
core ontologies of manufacturing. Usman et al. [22] forms the Manufacturing
Core Concepts Ontology (MCCO) by identifying core set of concepts formalized160
in the upper-level ontology that serve as foundation ontology to provide the first
stage of a common understanding before developing the domain specific concepts
of design and production.
In what follows, we will adopt this granular perspective in conceiving of
the PLC as having a scope that includes processes of design and development,165
manufacturing, usage, maintenance and disposal, as well as the information,
materials, qualities, and functions that participate in these processes.
In particular, we take into account also the following key areas of interest
regarding the PLC identified by Young et al. [3]:
� Information regarding products including product geometry,170
� Potential supply chain capability,
8
� Knowledge of what has been done in the past, and
� Potential legislation, catalog data, and standards that affect decision
making.
2.3. Previous Work on Additive Manufacturing Ontology175
Additive manufacturing (AM) is defined by the American Society for Testing
and Materials (ASTM) as: the process of joining materials to make objects from
3D model data, usually layer upon layer, as opposed to subtractive manufac-
turing methodologies [25]. AM is nowadays widely used in industrial product
development, and its ability to create almost any possible shape through a180
process of building up a product layer by layer.
Existing work on the ontology for AM includes SAMPro, for Semantic
Additive Manufacturing Process Planning, described by Eddy et al. [26]. SAMPro
is extended from the MSDL; it provides the starting point for a module that
includes types of AM such as the Binder Jetting and Directed Energy Deposition185
as its classes and focuses on the products which are the output of such processes,
representing in detail product features such as surface finish, accuracy, tolerance,
and so forth.
By contrast, The Design for Additive Manufacturing Ontology (DFAM),
presented in Dinar and Rosen [27], focuses on the detailed representation of190
different types of AM processes in terms of what it calls process parameters such
as printing orientation angle. However, DFAM, too, suffers from the fact that it
has been developed with its own peculiar vocabulary for the included classes.
NIST [28], Roh [29] and Liang [30] are some other ontologies for AM that
have been developed recently. Roh [29] develops an ontology for AM to represent195
information for different process models for laser, thermal, micro structure, and
mechanical properties for metal-based AM of Ti-6-6Al-4V. Liang [30] develops
the AM-OntoProc ontology that promote the modeling and reutilization of
9
knowledge towards the AM process planning where AM process is supposed to
begin from the utilization of CAD software during the design stage until the200
final AM prototype is developed.
Finally, there is recent work by Hagedorn et al. [31], which uses BFO as
the platform for an AM ontology called Innovative Capabilities of Additive
Manufacturing (ICAM). ICAM also reuses the BFO-conformant ontology - the
Information Artifact Ontology (IAO) - to provide the higher-level representation205
of information-related types that serves as its backbone. The information in
ICAM covers basic product attributes from the NIST Core Product Model
(CPM) relating to materials, geometry and designed function as well as types of
manufacturing processes and services taken over from MSDL. It also incorporates
the SAMPro model of AM and a set of formal description of parts and features210
from Functional Basis Ontology (FBO). ICAM is thus able to provide extensive
coverage of the AM domain and since our version of AMO also reuses the BFO-
conformant ontology - the Common Core Ontologies (CCO), we will be looking
forward for the opportunity to develop future versions of AMO to be consistent
with the ICAM content that focus on the application of AM. The CCO is a215
suite of ontologies that was released to the public recently that adds general
contents to the BFO structure and at the same time are also common to many
domain of interest, especially to manufacturing engineering domain. We feel
that Information Entity Ontology (IEO) that is part of CCO seems to be able to
represent information-related types to manufacturing domain in a more accurate220
way than IAO. However, there are still works to be done in making the IEO and
IAO to be compatible to each other.
10
3. Approach for Ontology Development
Although there are is no standard methodology for developing ontologies,
Natalya F and Deborah L [32] outlined a simple knowledge-engineering method-225
ology in developing ontology which was followed as a guideline in our work. The
methodology include: determining ontology domain and scope, considering on-
tology reuse, enumerating important terms, defining classes and class hierarchies,
defining class properties, defining values for properties, and creating instances of
classes.230
We adopted a top-down approach in most of the ontology development
process where the AMO as the domain ontology was constructed by downward
population from a common upper-level ontology in the multi-tiered network
connected in the following way [33]:
1. A single, small, domain-neutral upper-level ontology;235
2. Mid-level ontologies covering broad domains having root nodes that are
either direct children of classes from the upper-level ontology or of a term
drawn from another mid-level ontology within the network;
3. Lower-level ontologies representing specialized domains having root nodes
that are either direct children of classes from one of the mid-level ontologies240
or of a term drawn from another domain level ontology within the network;.
BFO and Descriptive Ontology for Linguistic and Cognitive Engineering
(DOLCE) are some of the upper-level ontology that have been used as the
foundation ontologies in the domain ontology development [34, 35, 36]. Both
ontologies in fact grew out of a common philosophical orientation, and thus some245
parts of the ontologies overlapped with each other. Since our work is part of the
Coordinated Holistic Alignment of Manufacturing Processes (CHAMP) project
founded by Digital Manufacturing and Design Institute (DMDII) which focuses
on constructing an efficient scheme to manage discordant manufacturing data
11
source, BFO is already selected as one of the project requirement. However,250
despite of the selection of BFO due to the CHAMP project requirement, BFO’s
well-documented guidelines and training material, it’s extensive use in in hundreds
of projects in biomedical and military domains, and increasingly being adopted
in industry as a top-level framework are another factors that contribute to the
selection of BFO. ICAM [31], CCO [37] and Functional Graded Material Ontology255
(FGMO)[38] are some of the ontologies that have been developed by adopting
BFO as foundation of the ontology development. These factors make its use a
key enabler in promoting the secondary use of our ontology by others.Therefore,
we selected BFO1 as an upper-level ontology to serve as starting point in the
ontology development of AMO. In addition, we have wherever possible reused260
content taken from the Common Core Ontologies (CCO)2 , which were also built
as conservative extensions of BFO.
BFO contains the top class entity that contains two subclasses: continuant
and occurrent [34]. A continuant is an entity that is wholly present at every
time during the course of its existence. Examples of continuant entities include265
objects, such as tables and people, as well as spatial regions and portions of
matter, qualities, such as the length of an airplane wing, and dispositions, such
as the tensile strength of a steel sheet. An occurrent, by contrast, is an entity
that occurs or happens by unfolding itself through time in successive phases (for
example of a beginning, middle, and end). Manufacturing processes fall under270
this heading, as also do the temporal regions during which such processes occur.
A fragment of the BFO class hierarchy is provided in Figure 2.
An AM process is a process that involves certain sorts of material entities
as its participants. A Portion of Material is a subclass of material entity, and
1See https://github.com/BFO-ontology/BFO/blob/master/bfo.owl2See https://github.com/CommonCoreOntology/CommonCoreOntologies
12
Figure 2: A fragment of the BFO class hierarchy [39].
instances of Portion of Material are the inputs for instances of AM process.275
Both processes and material entities are distinct from the third type of
entity in BFO comprising generically dependent continuants. These are, roughly,
patterns that can be exactly copied - they are entities that depend on the
existence of at least one bearer at any time during which they exist, but not on
any particular bearer. The most important subclass of generically dependent280
continuant is Information Content Entity, whose instances stand in a relation of
aboutness to some entity [40]. Importantly, for our purposes, this class includes
instances of reports, sentences, and data values that are about the processes and
materials of a manufacturing process. Such information is not itself material,
nor is it a process, though it may participate in processes [34].285
The Common Core Ontologies (CCO) form a set of conservative extensions
of BFO with the goal of representing the mid-level entities involving agents,
artifacts, actions, and measurements [41]. The ontologies in the CCO include:
13
� Agent Ontology, representing agents, especially persons and organizations,
and their roles.290
� Artifact Ontology, representing deliberately created material entities along
with their models, specifications, and functions.
� Currency Unit Ontology, representing currencies in different countries.
� Extended Relation Ontology, representing relations (i.e. object properties)
holding between entities.295
� Event Ontology, representing processes.
� Geospatial Ontology, representing sites, spatial regions, and other entities,
especially those that are located near the surface of Earth, as well as the
relations that hold between them.
� Information Entity Ontology, representing generic types of information as300
well as the relationships between information and other entities.
� Quality Ontology, representing a range of attributes of entities, including
qualities, realizable entities such as dispositions and roles, and process
profiles.
� Time Ontology representing temporal regions and the relations that hold305
between them. A temporal region, as defined by BFO, is an occurrent
entity that is part of time as defined relative to some reference frame.
� Units of Measure Ontology, representing standard units used when mea-
suring various attributes of entities.
Figure 3 shows the CCO ontologies with the import structure between them.310
Every class in CCO is the subclass of some class in BFO, and general relations
used in BFO are also adopted by the CCO.
14
Figure 3: The Common Core Ontology import structure [37].
Although CCO served as the mid-level ontology that reduce the generality
of BFO, having direct children from CCOs’ classes to represent the specific
domain in AM are not sufficient. There is a need of another level of ontology315
after the CCO so that it can increase the granularity of domain before moving
towards the concrete entities that represent AM. The Coordinated Holistic
Alignment of Manufacturing Processes (CHAMP) is a project funded by the
Digital Manufacturing and Design Innovation Institute (DMDII). The CHAMP
project has developed a suite of ontologies whose objective is to aid industrial320
organizations in overcoming the problem of data heterogeneity. The CHAMP3
ontologies are an extension of CCO representing the mid-level classes relating
to the design, manufacturing, use, and maintenance phases of the PLC. These
ontologies include:
3See https://github.com/NCOR-US/CHAMP
15
� Product Life Cycle Ontology325
� Commercial Entities Ontology
� Design Ontology
� Manufacturing Process Ontology
� Testing Process Ontology
� Tool Ontology330
� Maintenance Ontology
Figure 4 shows the CHAMP ontologies and their import structure.
Figure 4: The CHAMP Ontologies
Ideally, we planned to extend the whole set of CHAMP ontologies for the
development of AMO. However, since the CHAMP ontologies are still in initial
implementation, we decided to import only a fragment of the CHAMP ontologies335
that are related to the AM processes. We imported manually some of the
16
classes from the Design Ontology, Manufacturing Process Ontology, and Testing
Process Ontology without changing its’ class structures and URI from its’ original
ontologies. Figure 5 shows the framework for the AMO development.
Figure 5: An overview of AMO development.
4. The Additive Manufacturing Ontology (AMO)340
Existing classes from BFO, CCO, and CHAMP are imported into AMO and
new classes are added in a process of downward population. This ensures that
AMO utilizes commonly used terms and definitions and thereby increases the
chances that AMO will itself be re-used and integrated with other ontologies.
4.1. Process345
There is a canonical order to processes that occur within AM, and that
the processes represented in the AMO were selected in order to account for
this canonical representation. From a mechanical perspective, AM often uses
numerically controlled (NC) machines that are integrated with CAD and process
planning software. The canonical AM process flow consists of six steps [42]:350
1. 3D CAD model generation;
2. Conversion of the CAD Model into AM machine acceptable format (STL
file);
3. Setting the process parameters;
17
4. The process of printing;355
5. Support removal; and
6. Post-processing.
The class process is defined in BFO as an occurrent entity that exists in
time by occurring or happening, has temporal parts, and always depends on
some (at least one) material entity [34]. Each of the steps listed by Yang et al.360
[42] corresponds to the instantiation of a certain type of process entity. Process
entities in AMO include:
1. Design Ontology: ActOfDescribingClientNeed
2. Design Ontology: ActOfAnalysisOfClientNeed
3. AMO: ActOfCADModelDevelopment365
4. AMO: ActOfDataTransformation
5. Design Ontology: ActOfManufacturingRequirementIdentification
6. Manufacturing Process Ontology: ActOfAdditiveManufacturing
7. AMO: ActOfSupportRemoval
8. AMO: ActOfPostProcessing370
As can be seen, some of these classes are imported from the Design Ontology
and the Manufacturing Process Ontology. Figure 6 shows the taxonomy of the
process classes in AMO. IntentionalAct as can be seen from the figure is the
subclass of Act class where both Act and IntentionalAct are the CCO classes,
extension of Process class in BFO. The definition of both class as follows:375
� Act is a process in which at least one agent plays a causative role.
� IntentionalAct is an Act in which at least one agent plays a causative role
and which is prescribed by some Directive Information Content Entity
held by at least one of the Agents.
18
Figure 6: Process classes in AMO.
In addition to the eight main process classes, there are also conditions where380
main process classes at the instance level has other process as part of the main
19
process. However, it is not right to put that process class as the subclass of the
main process since this process part relations only hold in some cases and not
all the time. Figure 7 shows the example of some ActOfSupportRemoval that
has process part an ActOfMaterialRemoval and an ActOfPostProcessing that385
has an ActOfAbrading and an ActOfJoining as process parts.
Figure 7: Example for process parts of Act of Support Removal and Act of Post Processing.
20
4.2. Material Entity
BFO defines a material entity as an independent continuant that has some
portion of matter as part [34]. Three types of material entity are recognized by
BFO: object, fiat object part, and object aggregate [34]. BFO does not assert that390
all material entities fall under one or other of these headings. Thus, portions
of liquid, gas, and plasma are classified (in the current version of BFO) as
immediate descendants of material entity.
Resources involved in the AM process can all be classified as material entities
in the AMO. These resources are:395
� Portion of Material
� 3D Printing Machine
� Printed Object
� Finished Object
Portion of Material in AMO is a direct child of material entity because it400
may describe those material entities that are not object aggregates. Material
entity classes in AMO are shown in Figure 8. Meanwhile, 3DPrintingMachine,
PrintedObject, and FinishedObject are types of object in BFO [43]. PrintedObject
and FinishedObject are defined classes with definitions as follows:
� PrintedObject is an object that is and output of some Act of Additive405
Manufacturing.
� FinishedObject is an object that is an output of some Act of Post-Processing.
4.3. Information Content Entity
An Information Content Entity as defined by the CCO is ”a Generically
Dependent Continuant that generically depends on some Information Bearing410
21
Figure 8: Material Entity classes in AMO.
Entity and stands in the relation of aboutness to some entity4”. The CCO also
provides three sub-relations of ‘is about ’, including: describing, prescribing, and
designating.
The AMO makes use of these distinctions provided in the CCO to provide
a series of classes of information that either prescribe, designate, or describe415
various AMO processes and resources. These are displayed in Figure 9 and newly
added classes appear in bold. CADSoftwareProgram, CADModel, STLDataFile,
and TechnicalDrawing are InformationContentEntity in AMO. Thus, all are
generically dependent continuants in BFO. A CADModel, for example, contains
information pertaining to the qualities that must inhere in a solid model of the420
sort that is represented in an ArtifactModel. An ArtifactModel in the CCO is a
subclass of ArtifactDesign. Both classes are defined in the CCO5 as follows:
� ArtifactDesign is a Directive Information Content Entity that is a specifi-
cation of an object, manifested by an agent, intended to accomplish goals,
in a particular environment, using a set of primitive components, satisfying425
a set of requirements, subject to constraints.
4See https://github.com/CommonCoreOntology/CommonCoreOntologies/blob/master/
AllCoreOntology.ttl5See https://github.com/CommonCoreOntology/CommonCoreOntologies/blob/master/
AllCoreOntology.ttl
22
� ArtifactModel is an Artifact Design that prescribes a common set of
functions and qualities that are to inhere in a set of artifact instances.
Figure 9: Information Content Entity classes in AMO.
23
The CCO class, Directive Information Content Entity is defined as an Infor-
mation Content Entity that prescribes some entity. Specifications, for example,430
are classified under Directive Information Content Entity, as for example where
some Additive Manufacturing Process Specification prescribes some Act of Addi-
tive Manufacturing.
Requirements, too, are classified under Directive Information Content Entity,
for example when a Customer Requirement prescribes certain qualities that must435
inhere in a product. Standards documents such as ISO 3923, the International
Standard for Metallic Powders, prescribes the level of quality that a metallic
product must have if it is to satisfy the standard.
4.4. Relations
Figure 10 provides an overview of the relations used in AMO, in addition to440
process part of discussed in Section 4.1.
Definitions from BFO and the CCO are as follows:
� has participant is a primitive instance-level relation between a process, a
continuant, and a time at which the continuant participates in some way
in the process.445
� is input of is a relation between a continuant and a process in which the
continuant participates. The presence of the continuant at the beginning
of the process is a necessary condition for the initiation of the process.
� is output of is a relation between a continuant and a process in which the
continuant participates. The presence of the continuant at the end of the450
process is a necessary condition for the completion of the process.
� prescribes is for all types T1 and T2, if T1 prescribes T2, then there is
some instance of T1, t1, that serves as a rule or guide to some instance of
T2, t2.
24
Figure 10: An overview of AMO entities and relations.
25
5. Application of AMO to the Dentistry Product Manufacturing455
AMO was developed to serve as a mid-level ontology that can be re-utilized
for multiple different types of additive manufacturing. By providing terms for all
processes within the AM organized process flow, we feel that AMO is particularly
well suited for PLC applications. Besides, to ensure generality of AMO, the
properties interrelating objects of different ontologies are only defined directly460
in the AMO at the instance level. To show the utility of the developed AMO
for developing application-specific ontologies, a case study applying AMO to
dentistry manufacturing application termed Additive Manufacturing for Dental
Product Ontology (AMDO) is discussed next.
5.1. Additive Manufacturing (AM) in Dentistry465
AM has established itself in the dentistry field as a promising alternative to
the conventional manufacturing processes. AM has the advantage of yielding
accurate one-off fabrication of complex structures in a variety of materials
having properties highly desirable for both dentistry and surgery [44]. AM
even becoming a feature of many dental surgeries [45], where it allows direct470
fabrication of dental prostheses such as crowns and bridges.
As a case study, an application ontology extending AMO with new classes
related to the dentistry product manufacturing field has been developed. This
application ontology is titled the Additive Manufacturing for Dental Product
Ontology (AMDO) and is depicted in Figure 11 (newly added classes are high-475
lighted in bold). As can be seen from the figure, the newly added classes are the
extension of the existing classes from the imported ontologies. This shows the
re-usability of AMO in the developing the application ontology. No new object
properties are needed to be created as well.
26
Figure 11: Active classes in AMDO.
5.2. Case Demonstration480
This section demonstrate an example of the practical uses of the AMDO
guided by the work outlined in Khalil et al. [46]. In their work, they evaluate
dimensional differences between natural teeth and the printed models using three
different AM processes. The process starts with the scanning of three premolars
27
dimension from two dry adult human mandibles by means of an optical scanner485
to yield the three-dimensional data needed for the dental model. Table 2 shows
the specifications of the four 3D printers used in the study Khalil et al. [46].
Table 2: Specification of 3D Printers used
Machine SLA Objet Eden 250 Objet Connex 350 UP Plus 2
Printing type SLA Polyjet Polyjet FDMLayer thickness (µm) 50 16 16 150Material used Resin Resin Resin ABS
24 printed premolar tooth were produced in total and the volume of each
replicas are measured and compared against the original premolars. Table 3
shows the overview of the volume measurements of each printed premolar tooth490
with the percentage of volume difference with the original premolars.
Table 3: Volume and Percentage of Volume Differences DataGroup Tooth No. UP Plus 2 Objet Connex 350 Objet Eden 250 SLA
M1 34 509.2(4.2%) 502.1(2.7%) 482.6(-1.3%) 484.2(-0.9%)M1 44 616.7(18.5%) 521.2(0.1%) 516.2(-0.9%) 519.3(-0.3%)M1 54 440.5(-1.6%) 438.8(-2.0%) 446.2(-0.3%) 448.0(0.1%)M2 34 376.5(-4.4%) 421.6(7.0%) 398.2(1.1%) 388.9(-1.3%)M2 44 340.2(-6.8%) 351.0(-3.8%) 361.0(%-1.1%) 362.1(-0.8%)M2 54 416.4(-12%) 464.2(-1.9%) 482.6(%-0.9%) 471.1(-0.4%)
For the testing purposes of AMDO, we populated all data from Table 2 and
Table 3 as instances for the classes in AMDO. Following are the three queries
that we have made for the AMDO to provide the inferred information:
1. What are the 3D Printers used in the study?495
2. What type of materials used for each machine and what are the layer
thickness specification in fabricating the printed tooth?
3. What are volume differences of the printed tooth with the uses of different
3D printers?
The queries have been created using an Resource Description Framework (RDF)500
query language, SPARQL that is a plugin to Protege. SPARQL is a semantic
query language that is able to retrieve and manipulate data stored in the
28
RDF format where the entities and its relations are expressed in the form of
subjectpredicateobject. For each queries, following are the namespace and its
binded prefixes that were used to identify the URI of the classes:505
rdf: http://www.w3.org/1999/02/22-rdf-syntax-ns#
owl: http://www.w3.org/2002/07/owl#
rdfs: http://www.w3.org/2000/01/rdf-schema#
xsd: http://www.w3.org/2001/XMLSchema#
bfo: http://purl.obolibrary.org/obo/510
ros: http://www.obofoundry.org/ro/ro.owl#
ccos: http://www.ontologyrepository.com/CommonCoreOntologies/
plco: http://www.semanticweb.org/no/ontologies/2017/1/PLC-ontology/
mpos: http://www.semanticweb.org/ontologies/ManufacturingProcessOntolog
y/515
dos: http://www.semanticweb.org/no/ontologies/2017/6/DesignOntology/
amos: http://www.semanticweb.org/munira/ontologies/2017/6/AdditiveMan
ufacturingOntology#
amdo: http://www.semanticweb.org/munira/ontologies/2018/1/AdditiveMan
ufacturingDentalOntology#520
Query 1: Identifying 3D printers used in the study.
The SPARQL Query for this question is as follow:
SELECT ?Machine
WHERE {525
?Machine rdf:type amos:3DPrintingMachine }
As can be seen from Table 2, there are four 3D printers used in the study. Even
though there are only four data, this simple query represents those with large
29
input and low selectivity and does not assume any hierarchy information or530
inference. Figure 12 shows the result of this query.
Figure 12: Query 1 Result
Query 2: Identifying type of materials used for each machine and what are the
layer thickness specification in fabricating the printed tooth.
To identify the type of materials used for each machine and with the layer535
thickness specification for the printing process, the relations between the process,
machine, materials and also the process specification are defined at the instance
level. To increase the selectivity of the inferred information, we limited type of
printing process to only to the Vat Photopolymerisation Process. The SPARQL
Query for this question is as follow:540
SELECT ?Machine ?Material ?LayerThickness ?MeasurementUnit
WHERE {
?a rdf:type amos:ActOfVATPhotopolymerisation .
?Machine ros:agent in ?a .
?Machine rdf:type amos:3DPrintingMachine .545
?b ccos:is input of ?a .
?b rdf:type amos:PortionOfMaterial .
?b ccos:has text value ?Material .
?c ccos:prescribes ?a .
?c rdf:type amos:LayerThicknessSpecification .550
?c ccos:inheres in ?d .
?d rdf:type ccos:InformationBearingEntity .
30
?d ccos:has integer value ?LayerThickness .
?d ccos:uses measurement unit ?MeasurementUnit }
This query increases in complexity where there are four classes are involved and555
it has high selectivity due to the contraint added to one of the class. Figure 13
shows the result of this query.
Figure 13: Query 2 Result
Query 3: Identifying volume differences of the printed tooth with the uses of
different 3D printers?560
To identify the volume differences of the printed tooth with the uses of different
3D printers,the relations between the process, machine, and also the volume
measurements with the volume difference analysis data are defined at the in-
stance level. The illustration of the relations between instances of each class is
shown in Figure 14.565
To increase the selectivity of the inferred information, we limited type of printing
process to only to the Vat Photopolymerisation Process. The SPARQL Query
for this question is as follow:
570
SELECT?Machine?ProstheticTooth?ProstheticToothVolume?VolumeDifference
Percentage
WHERE {
?a rdf:type amos:ActOfMaterialJetting .
?Machine ros:agent in ?a .575
31
Figure 14: Instances in AMDO
?Machine rdf:type amos:3DPrintingMachine .
?ProstheticTooth ccos:is output of ?a .
?ProstheticTooth rdf:type amdo:PremolarProstheticTooth .
?ProstheticTooth ccos:bearer of ?d .
?d rdf:type amos:VolumeCapacity .580
?d ccos:is measured by ?e .
?e rdf:type amos:QualityMeasurementInformationContentEntity .
?e ccos:inheres in ?f .
?f rdf:type ccos:InformationBearingEntity .
?f ccos:has decimal value ?ProstheticToothVolume .585
?e ccos:is input of ?g .
?g rdf:type amos:ActOfAnalysisOfInspectionData .
32
?h ccos:is output of ?g .
?h rdf:type amos:AnalysisMeasurementInformationContentEntity .
?h ccos:inheres in ?i .590
?i rdf:type ccos:InformationBearingEntity .
?i ccos:has decimal value ?VolumeDifferencePercentage }
Figure 15 shows the result of this query.
Figure 15: Query 3 Result
The answers to the example of queries show the ability of the ontology to595
retrieve information that matched to the queries even though the classes in
AMDO are imported from different ontologies. This is because, AMDO is an
extension of AMO and AMO is an extension of CHAMP which are extended
from the CCO and BFO. We use the import process in developing the ontology
to maintain the URI of the classes so that the naming of the class, the class600
structure, the class definition and the class relations will be standardized in all
related ontologies. This will ensure re-usability of the ontology. Even though,
we have not tested the interoperability of the ontology yet, but aiming for the
re-usability of the ontology is a starting point in achieving interoperability of
the ontology.605
Nevertheless, due to the import structure of the AMDO with the AMO,CHAMP,
CCO and BFO, we will have CCO terms available in AMDO. The CCO terms
33
may then extend the reach of AMDO to cope with corresponding data concern-
ing persons and organizations, roles of persons (for instance dental technician,
patient), measurement units, and cost factors. Thus, the functionality of AMDO610
can be extended to cope with a digital record or a patients history record in such
a way as to document the process of maintenance of the dental crown, comparing
susceptible of wear of the crown and of associated dental disorders in different
patients, perhaps incorporating also terms from the OBOFoundry6.
6. Conclusion and Future Work615
The AMO was developed within the context of a more general treatment
of the PLC. It will be helpful to users who employ AM in their work, and who
face the challenge of data integration faced by most modern industries today.
It can assist the designer in designing a new product, by enabling access to
bodies of data across the entire dentistry product manufacturing domain, for620
example relating to materials used, patient experiences, maintenance costs, and
so forth. The framework is also sufficiently general that it may accommodate
the generation of more fine-grained application ontologies in other areas where
AM technology is applied.
As the manufacturing industry is evolving rapidly and becoming more com-625
petitive, quality and cost are major factors that need to be focused on by the
manufacturers. These factors are affected closely by the process and the material.
We concentrated here primarily on the process aspect in the AM process but
in the next stage, we will work on integrating the AMO with the ontology that
represents the types of material used in AM and their associated attributes.630
This will build on work on the ontology of material that is part of the CHAMP
ontologies, where each ontology in CHAMP constitutes a mid-level ontology that
6See http://www.obofoundry.org/ontology/ohd.html
34
imports the whole of the CCO, as well as BFO.
References
[1] X. Chang, R. Rai, J. Terpenny, Development and utilization of ontologies in635
design for manufacturing, Journal of Mechanical Design 132 (2010) 021009.
[2] A. Gunasekaran, E. Ngai, Knowledge management in 21st century manufac-
turing, International Journal of Production Research 45 (2007) 2391–2418.
[3] R. Young, A. G. Gunendran, A.-F. Cutting-Decelle, M. Gruninger, Manu-
facturing knowledge sharing in plm: a progression towards the use of heavy640
weight ontologies, International Journal of Production Research 45 (2007)
1505–1519.
[4] T. Economist, Knowledge management in manufacturing, 2007. URL:
www.01net.it/wp-content/uploads/sites/14/2014/10/siemens_ugs_
Knowledge_management.pdf.645
[5] S. Lemaignan, A. Siadat, J.-Y. Dantan, A. Semenenko, Mason: A proposal
for an ontology of manufacturing domain, in: Distributed Intelligent Systems:
Collective Intelligence and Its Applications, 2006. DIS 2006. IEEE Workshop
on, IEEE, 2006, pp. 195–200.
[6] C. Welty, D. L. McGuinness, M. K. Smith, Owl web ontology language650
guide, W3C recommendation, W3C (February 2004) http://www. w3.
org/TR/2004/REC-owl-guide-20040210 (2004).
[7] B. Motik, P. F. Patel-Schneider, B. Parsia, C. Bock, A. Fokoue, P. Haase,
R. Hoekstra, I. Horrocks, A. Ruttenberg, U. Sattler, Owl 2 web ontol-
ogy language: Structural specification and functional-style syntax, W3C655
recommendation 27 (2009) 159.
35
[8] T. Kjellberg, A. von Euler-Chelpin, M. Hedlind, M. Lundgren, G. Sivard,
D. Chen, The machine tool modela core part of the digital factory, CIRP
annals 58 (2009) 425–428.
[9] M. A. Musen, Upper Level Ontology. In: Dubitzky, W., Wolkenhauer,660
O., Yokota, H., Cho, K-H.(eds)Encyclopedia of systems biology, Springer
Publishing Company, Incorporated, 2013.
[10] Ramos, 2011. URL: https://downloads.sourceforge.net/project/
ontomanufacture/mop/Machine_Ontology.owl.
[11] L. Ramos, R. Gil, D. Anastasiou, M. J. Martin-Bautista, Towards a machine665
of a process (mop) ontology to facilitate e-commerce of industrial machinery,
Computers in Industry 65 (2014) 108–115.
[12] L. Mazzola, P. Kapahnke, M. Vujic, M. Klusch, Cdm-core: A manufacturing
domain ontology in owl2 for production and maintenance, in: KEOD, ????,
pp. 136–143.670
[13] F. Ameri, C. Urbanovsky, C. McArthur, A systematic approach to develop-
ing ontologies for manufacturing service modeling, in: Proc. 7th Interna-
tional Conference on Formal Ontology in Information Systems (FOIS 2012),
Graz, Austria, Citeseer, 2012.
[14] L. Mesmer, A. Olewnik, Development of a part-focused manufacturing pro-675
cess ontology: Exploring use and applications, in: ASME 2015 International
Design Engineering Technical Conferences and Computers and Information
in Engineering Conference, American Society of Mechanical Engineers, 2015,
pp. V01BT02A034–V01BT02A034.
[15] B. Smith, 2017. URL: https://emmc.info/wp-content/uploads/2017/680
12/EMMC-IntOp2017-Cambridge_Smith_Buffalo.pdf.
36
[16] Y.-J. Chen, Y.-M. Chen, H.-C. Chu, Development of a mechanism for
ontology-based product lifecycle knowledge integration, Expert Systems
with Applications 36 (2009) 2759–2779.
[17] M. B. Raza, R. Harrison, Ontological knowledge based system for product,685
process and resource relationships in automotive industry, in: Proceedings
of the 1st International Workshop on Ontology and Semantic Web for
Manufacturing, Heraklion, Crete, Greece, ????, pp. 23–36.
[18] M. Borsato, Bridging the gap between product lifecycle management and
sustainability in manufacturing through ontology building, Computers in690
Industry 65 (2014) 258–269.
[19] H. Cao, P. Folan, Product life cycle: the evolution of a paradigm and
literature review from 1950-2009, Production Planning & Control 23 (2012)
641–662.
[20] A. Matsokis, D. Kiritsis, An ontology-based approach for product lifecycle695
management, Computers in industry 61 (2010) 787–797.
[21] N. Chungoora, R. Young, G. Gunendran, Z. Usman, N. Anjum, C. Palmer,
J. Harding, K. Case, A. CuttingDecelle, A model driven ontologybased
approach for manufacturing knowledge sharing in plm, Enterprise Interop-
erability: IESA’12 Proceedings (2012) 197–204.700
[22] Z. Usman, R. I. M. Young, N. Chungoora, C. Palmer, K. Case, J. Harding, A
manufacturing core concepts ontology for product lifecycle interoperability,
in: International IFIP Working Conference on Enterprise Interoperability,
Springer, ????, pp. 5–18.
[23] E. Urwin, Deliverable D3.2: Initial formal ontologies, Report,705
37
2014. URL: http://www.flexinet-fof.eu/Documents/FLEXINET%20D3.
2%20Initial%20Formal%20Ontologies.pdf.
[24] E. Urwin, C. Palmer, B. Young, E. Palacios, J. M. Pinazo, A. Font, R. Al-
marcha, Deliverable D3.5: Product-service-production ontologies report, Re-
port, 2016. URL: http://www.flexinet-fof.eu/Documents/FLEXINET%710
20D3.5%20Product-Service-Production%20Ontologies.pdf.
[25] A. Standard, Iso/astm 52900: 2015 additive manufacturing-general
principles-terminology, ASTM F2792-10e1 (2012).
[26] D. Eddy, S. Krishnamurty, I. Grosse, M. Perham, J. Wileden, F. Ameri,
Knowledge management with an intelligent tool for additive manufacturing,715
in: ASME 2015 International Design Engineering Technical Conferences and
Computers and Information in Engineering Conference, American Society
of Mechanical Engineers, 2016, pp. V01AT02A023–V01AT02A023.
[27] M. Dinar, D. W. Rosen, A design for additive manufacturing ontology,
Journal of Computing and Information Science in Engineering 17 (2017)720
021013.
[28] NIST, Nist’s owl ontology for additive manufacturing, 2016. URL: https:
//github.com/iassouroko/AMontology.
[29] B.-M. Roh, Ontology-Based Laser Metamodels and Visualization for Metal-
Based Additive Manufacturing, Thesis, 2017.725
[30] J. S. Liang, An ontology-oriented knowledge methodology for process plan-
ning in additive layer manufacturing, Robotics and Computer-Integrated
Manufacturing 53 (2018) 28–44.
[31] T. J. Hagedorn, S. Krishnamurty, I. R. Grosse, A knowledge-based method
for innovative design for additive manufacturing supported by modular730
38
ontologies, Journal of Computing and Information Science in Engineering
18 (2018) 021009.
[32] N. Natalya F, M. Deborah L, Ontology development 101: A guide to creating
your first ontology, 2001.
[33] 2016. URL: https://github.com/CommonCoreOntology/735
CommonCoreOntologies/blob/master/Best%20Practices%20of%
20Ontology%20Development.pdf.
[34] R. Arp, B. Smith, A. D. Spear, Building ontologies with basic formal
ontology, Mit Press, 2015.
[35] V. Mascardi, V. Cord, P. Rosso, A comparison of upper ontologies, in:740
Woa, volume 2007, ????, pp. 55–64.
[36] S. Borgo, P. Leito, The role of foundational ontologies in manufacturing
domain applications, in: OTM Confederated International Conferences” On
the Move to Meaningful Internet Systems”, Springer, 2004, pp. 670–688.
[37] CUBRC, Common Core Ontologies for Data Integra-745
tion, 2017. URL: https://www.cubrc.org/index.php/
data-science-and-information-fusion/ontology.
[38] F. Furini, R. Rai, B. Smith, G. Colombo, V. Krovi, Development of a
manufacturing ontology for functionally graded materials, in: ASME 2016
International Design Engineering Technical Conferences and Computers and750
Information in Engineering Conference, American Society of Mechanical
Engineers, 2016, pp. V01BT02A030–V01BT02A030.
[39] B. Smith, Basic Formal Ontology 2.0 Specification and User’s Guide,
2015. URL: https://github.com/BFO-ontology/BFO/raw/master/docs/
bfo2-reference/BFO2-Reference.pdf.755
39
[40] W. Ceusters, B. Smith, Aboutness: Towards foundations for the information
artifact ontology, in: Proceedings of the Sixth International Conference on
Biomedical Ontology (ICBO), volume 1515, CEUR, 2015, pp. 1–5.
[41] CUBRC, An Overview of the Common Core Ontologies, 2017. URL:
https://github.com/CommonCoreOntology/CommonCoreOntologies/760
blob/master/An%20Overview%20of%20the%20Common%20Core%
20Ontologies%201.0.pdf.
[42] L. Yang, K. Hsu, B. Baughman, D. Godfrey, F. Medina, M. Menon,
S. Wiener, Additive Manufacturing of Metals: The Technology, Materials,
Design and Production, Springer, 2017.765
[43] B. Smith, On classifying material entities in basic formal ontology, in: Inter-
disciplinary Ontology: Proceedings of the Third Interdisciplinary Ontology
Meeting, Keio University Press, 2012, pp. 1–13.
[44] A. Barazanchi, K. C. Li, B. AlAmleh, K. Lyons, J. N. Waddell, Additive
technology: update on current materials and applications in dentistry,770
Journal of Prosthodontics 26 (2017) 156–163.
[45] A. Dawood, B. M. Marti, V. Sauret-Jackson, A. Darwood, 3d printing in
dentistry, British dental journal 219 (2015) 521.
[46] W. Khalil, M. EzEldeen, E. Van De Casteele, E. Shaheen, Y. Sun, M. Shah-
bazian, R. Olszewski, C. Politis, R. Jacobs, Validation of cone beam com-775
puted tomographybased tooth printing using different three-dimensional
printing technologies, Oral surgery, oral medicine, oral pathology and oral
radiology 121 (2016) 307–315.
40